This report determines if the classic Romhilt–Estes score would predict better if points for its components were determined using a Cox hazard model and if the Cornell voltage criteria should replace the original criteria. Of the 20,903 subjects, the mean age was 43 ± 10 years and 90.6% were men. The mean follow-up for the population was 17 years, with 881 cardiovascular deaths; they were tested from 1987 to 1999 and followed until 2013. The new score was created with multipliers based on the Cox hazards of its elements with age bracket and gender included. The Cornell criteria were analyzed individually using Cox hazards with and without adjustments for age, gender, and African-American ethnicity and subsequently incorporated into the new score for analysis. For the new score, all 7 components were significant predictors of cardiovascular mortality with gender producing the greatest hazard ratio (HR) and left axis deviation and QRS duration >110 ms producing the lowest. For the original Romhilt–Estes score, 367 patients (1.8%) met the “definite” cutoff and had an HR of 5.6 (95% confidence interval 4.3 to 7.1). For the new score, 208 patients (1.0%) met the “definite” left ventricular hypertrophy cutoff and had an HR of 13.6 (95% confidence interval 10.8 to 17.3). The Romhilt–Estes had an area under the curve of 0.63, whereas the new score and new score with Cornell voltage both had an area under the curve of 0.7. In conclusion, our modified Romhilt–Estes score with new multipliers and without voltage criteria outperformed the original score.